Index of every page on this site: llms.txt
All skills
sergebulaev avatar

/yt-description-writer

@f422cea
Run it yourself

Write a full YouTube video description under 5,000 characters with a first-150-character hook (the only part visible before Show more, and what shows in search and suggested), timestamped chapters, naturally placed keywords, links, and a clear call to action. Covers long-form and YouTube Shorts. Becomes the create-post content that publishes with the video. Use to write or rework a description. Not for the title (use yt-title-optimizer) or the spoken opening (use yt-hook-scripter).

Use this Skill: https://skilld.dev/gh/sergebulaev/youtube-skills/yt-description-writer

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

SKILL.md

≈127 tokens for metadata: the name and description. ≈1.3k when used: this file. ≈660 more on demand in 1 file.

Description uses 6.4% of example budget

Before choosing a Skill, your Agent reads its name and description. All available Skills share that space.

  • A shorter description leaves more room for other Skills. This entry exceeds our 1% size suggestion.

In our Claude Code example, all Skill names and descriptions share 8,000 characters. This Skill uses ≈510 characters, or 6.4%.

The 1% threshold is a size suggestion. Longer descriptions can still fit.

Your model, settings, and other Skills decide how much text your Agent can read.

Example settings and source

The example uses a 200k-token context and default Claude Code settings. The count includes the name, description, separators, and when_to_use when present. Codex also counts local file paths.

Skit's source and limits: Codex 0.160.1, Claude Code 2.1.292.

YouTube Description Writer

The description does three jobs: it hooks the reader in its first 150 characters, it feeds YouTube and Google search, and it organizes the video with chapters and links. This skill writes all three layers.

When to use

  • User says "write the description for this video"
  • User has a video and wants chapters, keywords, and links structured
  • User wants the first 150 chars optimized for the search/suggested snippet
  • User is about to upload and needs the content that ships with the video

The three layers of a description

Layer Chars Job
Hook first ~150 the only text visible before "Show more"; also the search/suggested snippet. Lead with the payoff or the question and the search phrase.
Body ~150 to 1,000 expand the value, restate the promise, place the main keywords naturally, link the key resource.
Furniture rest, up to 5,000 chapters, links, social, credits, hashtags. Skimmable, labeled.

Steps

Voice profile first (all drafts). If ../../references/voice-profile.md has filled: yes, load it and match the user's voice fingerprint, hard rules, and CTA/link style throughout. If it is not filled, mention once that yt-community-post-writer --mode profile can learn their voice from a few posts, then proceed with the generic voice rules.

Never invent the specifics. The rules below ask for a concrete number, a date and a named entity, because that is what separates a real post from a generated one. Take them from what the user actually said in this conversation. Do not invent a figure, a date, a client name or a result, and do not soften a vague claim into a plausible-looking number. If the user has nothing concrete for a beat, ask them once, and if they still have nothing, drop the claim rather than decorate it. A published invented number is a retraction; a missing one is only a weaker post.

  1. Gather inputs. The title (so the description complements it), the video's payoff, the main search phrase plus 3 to 5 secondary keywords, the chapter beats with timestamps, any links (lead magnet, gear, socials), and whether it is long-form or a Short. If a URL is given, parse it with lib/url_parser.py.
  2. Write the first 150 chars as a second hook. Lead with the payoff or the question, work the primary keyword in naturally, and make it read well as a standalone snippet. This is not "welcome back to the channel".
  3. Write the body. 2 to 4 short paragraphs that restate the promise and place the keywords without stuffing. Put the single most important link here.
  4. Add chapters. Timestamps starting at 0:00, each label short and descriptive. YouTube needs at least 3 chapters and a 0:00 start to enable the chapter UI. For a Short, skip chapters.
  5. Add the furniture. Links section, socials, a one-line CTA (subscribe for the specific reason, not "smash like"), and 3 to 5 relevant hashtags at the very end (the first 3 surface above the title).
  6. Humanizer pass and length check. Scrub 2026 AI vocabulary by paragraph density (one marker is fine; three in one paragraph means rewrite it), cap em dashes at about one per 100 words (replace the excess with a comma, colon or line break, never a period), strip reveal bridges and sincerity openers ("let me be honest"); never add a hedge or confession the creator did not write. Keep the whole thing under 5,000 chars. Confirm the first 150 read well truncated.
  7. Approval card. Show the description with a marked 150-char cutoff and the total char count.
  8. On approval. This description is the content passed to lib.publish(kind="video"|"short", draft_text=<description>, title=<title>, ...). The title travels separately in platformSettings.youtube.title.

Hard rules

Global voice rules: see root SKILL.md Voice rules. Additional skill-specific rules:

  • 5,000-char hard cap. The first 150 chars carry the hook and the search phrase.
  • The first line is never boilerplate ("Hi everyone, welcome back").
  • Chapters: at least 3, first one at 0:00, short labels. Long-form only.
  • Keywords appear in natural sentences, never as a comma-separated dump.
  • Hashtags: 3 to 5 max, at the end. More than 15 makes YouTube ignore all of them.
  • Every link is labeled so the reader knows what it is before clicking.

Anti-patterns (skill will refuse)

  • A boilerplate greeting in the first 150 chars.
  • A keyword dump or a tag wall pretending to be a description.
  • 15+ hashtags (YouTube strips them all).
  • Em dashes above the cap (more than about one per 100 words), or an em dash swapped for a period.
  • "Smash that like button and subscribe" dead phrasing.
  • Chapters that do not start at 0:00 (the UI will not render them).
  • Raw unlabeled links.

Resources

  • ../../references/algorithm-heuristics.md - the 150-char snippet, chapters, search signals
  • ../../references/voice-rules.md - the canonical voice rules
  • references/description-template.md - the full skeleton with chapters, links, hashtags

Related skills

  • yt-title-optimizer - the title the description complements
  • yt-hook-scripter - the spoken first 30 seconds (different from the written hook)
  • yt-content-planner - keyword and link strategy across a series

Source: SKILL.md on GitHub

No rule matched.

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No alerts1mo3 checks · Risk SAFE
  • Gen Agent Trust Hub1mo

    The skill facilitates the creation and publication of YouTube descriptions by processing user input and fetching data from external URLs. A security analysis identified a standard indirect prompt injection surface where data retrieved from external sources is processed without explicit sanitization or boundary markers before being used to generate content for publication.

  • Socket1mo

    No alerts

  • Snyk1mo

    Risk: LOW · No issues

Signed by skilld at f422cea. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub yesterday.

Activeupdated 2 months ago

README badge

README badge for sergebulaev/youtube-skills/yt-description-writer